KR100986765B1 - A method and apparatus for evaluating the objective reconstructed quality of compressed video - Google Patents

Abstract

The present invention relates to a quality evaluation apparatus and method for evaluating the quality of a compressed and restored image transmitted from a transmitting unit which encodes an image of a camera and transmits it through a network to a receiving unit which decodes the encoded image and displays it on a display. Receiving the output of the camera, and stored in the S frame memory in frame units, the image input to the display in the D frame memory in frame units, and sequentially D frame for each of the frames stored in the S frame memory By calculating the distortion value with the frames stored in the memory, the R frame is minimized. When the R frame is calculated, the end-to-end delay of PSNR and frame time unit is calculated through the frame difference between the S frame and the R frame. The frame rate is calculated by analyzing the number of skipped frames when there are periodically continuous frames whose distortion is below the threshold. In this way, PSNR, end-to-end delay, and frame rate of compressed, transmitted, and reconstructed images can be measured without using time synchronization and without the help of system developers.

  Frame memory, image encoding, decoding, video frame, image quality

Description

A method and apparatus for evaluating the objective reconstructed quality of compressed video}

1 is a block diagram of a video encoding and decoding system.

2 is a block diagram of an image quality evaluation apparatus according to an embodiment of the present invention.

3 is a flowchart of a method for evaluating image quality according to an embodiment of the present invention.

4 is a diagram illustrating the relationship of signals to help understanding the concept of the present invention.

The invention relates to an apparatus for evaluating the quality of an image, in particular, not in the image compression algorithm region (hereinafter referred to as the "compression region") but in the actual end-to-end, ie two analog regions between the camera output and the playback display input. The present invention relates to a method and apparatus for measuring an objective quality of a transmitted and reconstructed image.

In the conventional measurement of compressed image quality, information on a frame rate may be obtained by measuring peak-to-peak signal to noise ratio (PSNR) in a compressed region or by analyzing syntax of a compressed bitstream.

However, these methods could not be observed externally by third parties without knowledge of the information flow in the encoder and decoder or without the help of the encoder and decoder developers.

Even if you measure the PSNR value in the compression domain with the help of the developer, the amount is strictly comparing the image quality between the input end of the encoder and the output end of the decoder. none.

The end-to-end delay cannot be measured without time synchronization between the sender and receiver, and even if it can be measured, it is impossible without the help of the system developer.

If hardware and software encoders and decoders are implemented by various manufacturers and need to measure image quality thereof, there is currently no quality evaluation method except subjective image quality measurement method.

The subjective measurement method is a method in which several testers take an image sequence and take an average based on the results of subjectively evaluating image quality.

Therefore, it is difficult to obtain an objective result and it is difficult to verify the reliability of the result because the evaluation result can be changed by various variables such as the emotion, preference, mental and physical state of the investigator.

An object of the present invention is to provide a method and apparatus for measuring PSNR, end-to-end delay, and frame rate of compressed, transmitted, and reconstructed video without using time synchronization and without the help of system developers. It is.

In the present invention, a true end-to-end delay including frame rate, network delay, and PSNR (peak-to-peak signal to noise ratio) of the reconstructed image as a quality measure of the reconstructed image. By considering three things and quantitatively calculating these values, we provide a method for evaluating the end-to-end quality of a decompressed image and an apparatus for implementing the method.

In order to solve this problem, an apparatus for evaluating quantitative quality of a decompressed image according to an aspect of the present invention,

A quality evaluation apparatus for evaluating the quality of a compressed restored image transmitted from a transmitting unit which encodes an image of a camera and transmits it through a network to a receiving unit which decodes the encoded image and displays it on a display.

A video frame capture unit which receives the output of the camera and stores the S frame memory in frame units, stores the image input to the display in D frame memory in frame units, and outputs the image according to an external request;

For each of the frames stored in the S frame memory, a distortion value with the frames stored in the D frame memory is sequentially calculated to calculate an R frame that minimizes the value.If the R frame is calculated, a frame difference between the S frame and the R frame is calculated. An analysis algorithm processing unit calculates the frame rate by calculating the end-to-end delay of PSNR and frame time unit and analyzing the number of skipped frames when there are periodic frames with distortion values below the threshold. It includes.

The quantitative quality evaluation method for the decompressed image according to one feature of the present invention for solving this problem,

A quality evaluation method for evaluating the quality of a compressed restored image transmitted from a transmitting unit transmitting an image of a camera through a network to a receiving unit decoding a coded image and displaying the same on a display.

Receiving an output of the camera and storing the image in an S frame memory on a frame basis and storing the image input to the display in a D frame memory on a frame basis;

Calculating an R frame whose value is minimized by sequentially calculating distortion values with frames stored in the D frame memory for each of the frames stored in the S frame memory;

Calculating an end-to-end delay in units of frame time based on a frame difference between the S frame and the R frame when the R frame is calculated;

Calculating a frame rate by analyzing the number of skipped frames when there are periodically continuous frames whose distortion is below a threshold.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention. Like parts are designated by like reference numerals throughout the specification.

1 is a block diagram of a general video encoding and decoding system.

Referring to FIG. 1, a video encoding / decoding system includes a transmitter 5 and a receiver 9.

The camera output signal of the transmitter 5 is input to the image encoder 3 in digital form via the preprocessor 2. The image encoder 3 compresses the image, and the network matcher 4 packetizes the compressed bitstream and transmits it to the network.

The receiving unit 9 restores the compressed bitstream from the packet received by the network matching unit 4 and passes the bitstream to the image decoding unit 7. The image decoder 7 restores the image by decoding the bitstream, and the restored image is output to the display through the post processor. At this time, the camera output and the input of the display unit are analog signals.

2 is a block diagram of an image quality measuring system according to an exemplary embodiment of the present invention, and FIG. 3 is a diagram illustrating an operation algorithm of an image quality analyzer.

Referring to FIG. 2, the image quality measuring apparatus according to the embodiment of the present invention includes a video frame capture unit 100 and an analysis algorithm processing unit 200.

The A / D converter of the video frame capture unit 100 performs A / D (analog-to-digital) conversion of the output signal of the camera and the post processor, that is, the display input signal. The output of the A / D converter 110 is divided and stored one frame at a time from the start of a new frame on the basis of frame sync, the signal from the camera output is S (source) frame memory 120, The signal from the post processor output is in turn stored in the destination (D) frame memory 130.

The two digitally converted video signals are stored in the S frame memory 120 and the D frame memory 130 in frame units for a predetermined frame time (for example, 1 second) that can calculate end-to-end delay.

Thereafter, the image signals stored in the memories 120 and 130 are transmitted to the image quality analyzer 210 through the interface units 150 and 220 under the control of the controller 140.

Here, the video signal stored in the D frame memory 130 may be regarded as a video signal obtained by compressing, transmitting, and restoring the video signal stored in the S frame memory 120. However, as shown in FIG. Has a time difference.

The image quality analyzer 210 sequentially calculates a distortion value (eg, mean square error (MSE)) between the frames stored in the D frame memory 130 for each of the frames stored in the S frame memory 120. The frame is found to be minimized, and the frame is regarded as a reconstructed R frame corresponding to one frame of the corresponding S frame memory 120.

If this frame is found, the image quality analyzer 210 can obtain the end-to-end delay of the PSNR and frame time unit through the frame difference between the S frame and the R frame, and the frame whose distortion falls below a certain threshold is periodically If the frame rate is continuously present, it is determined that the frame rate is not normal, and the frame rate is calculated by analyzing the number of skipped frames.

The end-to-end delay, PSNR, and frame rate calculated in this way are output by the image quality analyzer 210 to the outside and displayed on a display or supplied to a source.

The operation of obtaining the end-to-end delay, PSNR, and frame rate will be described in detail as follows.

In FIG. 4, the camera output is S ₁ ,. And output the post-processor output to D _j . Expressed as

The actual frame that has been compressed, transmitted, or reconstructed for S ₁ will be one frame of the D _j frames, and depending on the image compression algorithm used, it may be a frame that is actually compressed, not transmitted, and skipped. Since the image quality of the image is very sensitive to the brightness component, it is assumed herein that only the brightness component of each frame is used for convenience.

The output image R _i corresponding to the arbitrary input image S _i is regarded as a frame satisfying the following equation 1 in D _j .

　　

　　

Here, f (x) is a distortion value between the D frame and the S frame, and is defined as f (X) = E [X ² ].

R _i R _i is obtained between the floor corresponding to the S _i that is the end-to-end delay for the i-th frame is DELTA i

It is obtained as in Equation 2.

Here, t (R _i ) means a time point at which the i-th frame is played and t (S _i ) means a time point at which the i-th frame is input.

모든 i 값에 대해 수학식 1과 수학식 2를 적용하면 평균 종단간 지연

Equation 1 and Equation 2 for all i values mean average end-to-end delay

The average picture quality (PSNR) of the reconstructed video is obtained as follows.

이다.here,

to be.

If there is a skipped frame in the encoding process, the receiver will repeatedly play the previous frame at the playback frame time of the frame. In this case, since the PSNR value between the input image and the reconstructed image drops sharply with respect to the missing frame, the missing frame can be detected.

If the i-1 th frame is normally encoded and decoded, it may be detected whether the i th frame is missing through Equation 5.

　

　

The frame rate can be obtained by subtracting the number of periodically deleted frames out of N frames.

If there is a skipped frame, the average value may be taken only for the actual encoded frames except for the missing frame when the average value is obtained in Equations 3 and 4.

However, the subjective picture quality of the dropped frame should be reflected in the quality evaluation, which is a matter of determining according to the quality measurement policy, which is beyond the scope of the present invention.

The implementation of the proposed method consists of a video frame capture unit for capturing video frames and an analysis algorithm processing unit for processing the image quality analysis method, which can be processed in one device. It can also be processed through a separate device. The invention includes both implementation methods.

Unlike conventional methods for evaluating the quality of an image, the present invention evaluates end-to-end quality between a camera output end and an input end of a final display device.

In addition, the end-to-end quality, which could not be achieved by the conventional method, is measured because the quality of the stage immediately before being expressed to the user is compared without evaluating the quality in the compressed region. In addition, the proposed method can measure not only the PSNR but also the amount of end-to-end delay and frame rate.

The present invention is not limited only to the above-described embodiments, but may be modified and modified within the scope not departing from the gist of the present invention. Those skilled in the art will appreciate that various changes and modifications can be made without departing from the spirit and scope of the present invention.

As described above, in the embodiment of the present invention, it is possible to provide an image quality measuring apparatus and method having the following effects.

1) It is possible to measure the quality of end-to-end image that is expressed directly to the user rather than the compression algorithm.

2) The objective quality of the playback image can be calculated regardless of the image compression algorithm used.

3) By taking image data in the analog domain and processing it, third parties can measure the quality without the help of system developers.

4) Quality information such as PSNR, end-to-end delay, and frame rate can be obtained simultaneously.

Claims (8)

delete A quality evaluation apparatus for evaluating the quality of a compressed restored image transmitted from a transmitting unit which encodes an image of a camera and transmits it through a network to a receiving unit which decodes the encoded image and displays it on a display. A video frame capture unit which receives the output of the camera and stores the S frame memory in frame units, stores the image input to the display in D frame memory in frame units, and outputs the image according to an external request; For each of the frames stored in the S frame memory, a distortion value with the frames stored in the D frame memory is sequentially calculated to calculate an R frame that minimizes the value.If the R frame is calculated, a frame difference between the S frame and the R frame is calculated. An analysis algorithm processing unit for calculating the end-to-end delay of the PSNR and the frame time unit, and calculates the frame rate by analyzing the number of missing frames when there are periodically continuous frames with a distortion value less than the threshold, The frame capture unit, An A / D converter configured to perform A / D conversion on the camera output signal and the display input signal, respectively; S frame memory for storing the camera output of the A / D conversion unit separately by one frame from the moment when a new frame starts based on frame sync), and the display input based on frame sync. A D frame memory for storing one frame at a time from the beginning of the frame; An interface unit to interface the S frame memory and the D frame memory to the analysis algorithm processor; And a controller configured to control the camera output and the display input to be stored and output to the analysis algorithm processor. The method of claim 2, And two video signals digitally converted by the A / D converter are stored in the S frame memory and the D frame memory on a frame-by-frame basis for a predetermined frame time to calculate end-to-end delay. The method according to claim 2 or 3, The analysis algorithm processing unit, A second interface unit for interfacing data received from the video frame capture unit; For each of the frames stored in the S frame memory, the distortion value with the frames stored in the D frame memory is sequentially calculated through the second interface unit to calculate an R frame whose value is minimized. Image quality analysis that calculates the frame rate by calculating the frame rate by calculating the end-to-end delay in units of frame time based on the frame difference between R and R frames, and when the frames whose distortion value is below the threshold periodically exist continuously. Quantitative quality evaluation apparatus comprising a part. delete A quality evaluation method for evaluating the quality of a compressed restored image transmitted from a transmitting unit transmitting an image of a camera through a network to a receiving unit decoding a coded image and displaying the same on a display. A first step of receiving an output of the camera and storing it in an S frame memory on a frame basis and storing the image input to the display in a D frame memory on a frame basis; A second step of calculating an R frame whose value is minimized by sequentially calculating a distortion value with the frames stored in the D frame memory for each of the frames stored in the S frame memory; Calculating an end-to-end delay in frame time units based on a frame difference between the S frame and the R frame when the R frame is calculated; A fourth step of calculating a frame rate by analyzing the number of skipped frames when there are periodically continuous frames whose distortion is below a threshold, The second step is a quantitative quality evaluation method performed using the following equation.

　　 Here, f (x) is a distortion value between the D frame and the S frame, and is defined as f (X) = E [X ² ]. A quality evaluation method for evaluating the quality of a compressed restored image transmitted from a transmitting unit transmitting an image of a camera through a network to a receiving unit decoding a coded image and displaying the same on a display. A first step of receiving an output of the camera and storing it in an S frame memory on a frame basis and storing the image input to the display in a D frame memory on a frame basis; A second step of calculating an R frame whose value is minimized by sequentially calculating a distortion value with the frames stored in the D frame memory for each of the frames stored in the S frame memory; Calculating an end-to-end delay in frame time units based on a frame difference between the S frame and the R frame when the R frame is calculated; A fourth step of calculating a frame rate by analyzing the number of skipped frames when there are periodically continuous frames whose distortion is below a threshold, In the third step, The end-to-end delay DELTA i is calculated by the following equation.

Here, t (R _i ) means a time point at which the i-th frame is played and t (S _i ) means a time point at which the i-th frame is input. The method of claim 7, wherein

The third step may further include calculating an average end-to-end delay and an average picture quality (PSNR) of the reconstructed image. The quantitative quality evaluation method of the average quality of the image and the image is obtained by the following equation.

here,

to be.

Images